Agentic Commerce: Why Identity Alone Will No Longer Be Enough
Market Analysis by Giesecke+Devrient
AI agents are increasingly initiating payment transactions without human involvement: reordering materials, autonomously booking and paying for hotel stays, and settling tolls in real time. Technically, all of this has long been feasible. What now moves into focus is the layer of trust: How do providers authenticate a transaction when the party initiating it is not a real person? G+D Netcetera puts the key questions into context.
Why do proven authentication models break down in agentic commerce?
Strong customer authentication, 3-D Secure, and biometric methods were designed for a human-in-the-loop approach, in which real people are aware of an action and actively approve it. Agentic commerce follows an entirely different logic. The key question shifts from “Is this the right person?” to “Is this the right intent?”. Identity verification is therefore complemented by delegated authorization and by verification of context and purpose.
Is the payments industry even prepared for this change?
The ecosystem is not fully mature yet, but development is moving quickly. E-commerce platforms are already testing autonomous customer journeys, while card networks have introduced their own specifications for agent-driven transactions, including the Visa Trusted Agent Protocol and Mastercard Agent Pay. Since mid-2026, the first live implementations have been up and running: Visa reports completed live agent transactions with European merchants, backed by more than 30 issuing banks. The regulatory side, however, remains open. Strong customer authentication under PSD2 presupposes that the paying person initiates the transaction personally. Transactions initiated by an agent within an authorized mandate do not yet fit neatly into the existing framework. Classic fraud signals such as device, behavior, or location lose their predictive value in this setting. New questions take their place: Is the agent’s identity known? Is the mandate valid? Does the purchasing pattern match the delegated intent? The Agent Payments Protocol (AP2) has since been handed over to the FIDO Alliance, where specification work for agent-initiated commerce is under way in a working group chaired by Mastercard and Visa. This makes the chain of proof cryptographically verifiable. The development is not being driven by AI providers alone, but by collaboration among payment security specialists, infrastructure providers, identity experts, and established issuers and PSPs.
What role can banks and issuers play in this market, and what do they need to prepare?
For financial institutions, agentic commerce is a strategic opportunity, because some entity has to sign the mandate that gives an agent room to act. The role of credential provider is a logical continuation of what banks have done for decades: verifying identities, managing authorizations, securing payments. This role comes with a responsibility for which no established practice exists yet. If an agent buys the wrong product within a formally valid mandate, for instance, it is not a case of fraud but a dispute over the scope of the mandate. Current chargeback rules do not clearly define who bears the resulting loss. A technological reset is not required. Established solutions such as tokenization, EMV 3-D Secure, and FIDO authentication can be extended with delegated authorization, continuous real-time risk assessment, and context-based decision logic. What matters is integrating these components into consistent, auditable processes.
“AI agents will fundamentally change payments,” says Ivan Vukelikj, Secure Digital Payments expert at G+D Netcetera. “With every delegated purchase decision, the question of how trust can be established between people, AI agents, and payment infrastructures grows in importance. Those who lay the groundwork today for authorized, traceable, and context-based transactions are laying the foundation for the next stage of digital payments.”
Related links:
- G+D Netcetera: Who Authorizes the Agent?
- G+D Netcetera: Trust and Security as the foundation for AI rise
Press contact
Giesecke+Devrient
Christoph Lang
Head of Corporate Brand Communications and Government Relations
Phone: +49 89 4119-2164
E-Mail: christoph.lang@gi-de.com
PR-COM GmbH
Jessica Zola
Account Manager
Phone: +49 89 59997-815
E-Mail: jessica.zola@pr-com.de